Exploring an ontology via text similarity: an experimental study

In this paper we consider the problem of retrieving the conceptsof an ontology that are most relevant to a given textual query.In our setting the concepts are associated with textual fragments, suchas labels, descriptions, and links to other relevant concepts. The maintask to be solved is the definition of a similarity measure between thesingle text of the query and the set of texts associated with an ontologyconcept. We experimentally study this problem on a particular scenariowith a socio-pedagogic domain ontology and Italian language texts. Weinvestigate how the basic cosine similarity measure on the bag-of-wordstext representations can be improved in three distinct ways by (i) takinginto account the context of the ontology nodes, (ii) using the linearcombination of various measures, and (iii) exploiting semantic resources.The experimental evaluation confirms the improvement of the presentedmethods upon the baseline. Beside discussing some issues to consider inapplying these methods, we point out some directions for further improvement.